A method for early warning of condensate pump failure based on state parameters

Abstract

The present invention relates to a condensate pump failure early warning method based on state parameters. The method first selects training data and test data from the state parameters representing the condensate pump system in the operation data of the SCADA system, and secondly utilizes the condensate pump based on LS-SVR The multiple state regression model fits the state parameters of the condensate pump system, obtains the regression value of each state parameter of the condensate pump system, and compares the obtained regression value with the actual value to obtain the deviation. Finally, the state is determined by setting the deviation dynamic threshold based on the confidence interval. parameters for early warning of failures. Compared with the prior art, the invention has the advantages of realizing early fault identification, improving maintenance efficiency and high practicability.

Description

technical field [0001] The invention relates to the technical field of safety assessment of condensate pumps, in particular to a fault warning method for condensate pumps based on state parameters. Background technique [0002] In power plant water pump operation, condensate pump is a relatively common type of water pump. The steam that has done work in the steam turbine enters the condenser through the exhaust pipe, and after being cooled by the cooling water, it becomes condensed water and concentrates in the hot water well. Since the steam-water system is an ideal balancing process, the condensate pump must run continuously to transport the condensate to the boiler heater for a series of heating. It can be seen that if the condensate pump fails during normal operation, the water level of the hot well of the condenser will rise and the copper pipe of the condenser will be submerged, so that the steam cannot condense, and the vacuum will drop sharply, which will cause a se...

AI Technical Summary

Problems solved by technology

However, the above three methods are all manual operations, which require high experience of technicians, cannot realize automatic early warning, and have low efficiency in identifying faults

In the power plant DCS (Distributed Control System, decentralized control system), the upper and lower alarm limits of the state parameters of the existing automatic fault warning technology are fixed. Once the equipment parameters exceed the alarm line, the condensate pump system has almost failed, and it is necessary to Downtime for maintenance; however, in fact most failures do not occur suddenly, there is a slow development process

The existing technology often chooses a fixed value to set, for example, it is set according to the mean value of the residual sequence or the residual threshold value multiplied by a coefficient on the basis of the mean value. If the value is set small, it is easy to cause false alarms. However, if the value is set too large, it will not be able to play an alarm role. This method is not clear enough and has no practical application effect.

In addition, in terms of maintenance of condensate pumps, most power plants adopt maintenance methods such as planned maintenance and post-event maintenance, which have low maintenance efficiency and cannot guarantee the safety, reliability and utilization of condensate pumps.

Images